In [1]:
import pandas as pd
import geopandas as gpd
import movingpandas as mpd
from shapely.geometry import Point
from datetime import datetime, timedelta
import holoviews as hv
import hvplot.pandas 
import warnings
warnings.filterwarnings('ignore')
In [2]:
df = pd.read_pickle("data/interim/df_bike.pickle")
df
Out[2]:
id class frame_in frame_out xs_640x360 ys_640x360 num_frames time_on_screen_s x_start_640x360 x_end_640x360 y_start_640x360 y_end_640x360
0 2 Bicycle 250 409 [483.545, 483.7, 483.85, 484.0, 484.15, 484.3,... [181.87, 181.93, 181.985, 182.045, 182.1, 182.... 160 5.333333 483.545 632.195 181.870 103.390
3 8 Bicycle 250 398 [205.965, 205.775, 205.58, 205.39, 205.195, 20... [230.0, 229.27, 228.53, 227.79, 227.045, 226.3... 149 4.966667 205.965 301.880 230.000 6.935
4 10 Bicycle 257 423 [606.62, 602.955, 599.285, 595.615, 591.96, 58... [145.22, 146.265, 147.325, 148.425, 149.57, 15... 167 5.566667 606.620 297.715 145.220 280.260
8 18 Bicycle 523 1482 [194.58, 194.52, 194.455, 194.39, 194.325, 194... [275.42, 275.18, 274.94, 274.7, 274.455, 274.2... 960 32.000000 194.580 609.115 275.420 121.685
9 20 Bicycle 674 1655 [2.97, 3.425, 3.905, 4.395, 4.885, 5.375, 5.84... [193.985, 193.64, 193.275, 192.905, 192.53, 19... 982 32.733333 2.970 625.260 193.985 112.135
... ... ... ... ... ... ... ... ... ... ... ... ...
18837 37676 Bicycle 1096773 1096989 [625.37, 622.415, 619.44, 616.455, 613.465, 61... [138.795, 140.055, 141.325, 142.605, 143.895, ... 217 7.233333 625.370 307.340 138.795 276.935
18838 37678 Bicycle 1096841 1096918 [545.28, 543.88, 542.495, 541.115, 539.74, 538... [187.275, 187.81, 188.34, 188.87, 189.39, 189.... 78 2.600000 545.280 456.185 187.275 231.440
18839 37680 Bicycle 1096957 1097321 [637.23, 636.435, 635.65, 634.87, 634.095, 633... [137.145, 137.11, 137.085, 137.075, 137.095, 1... 365 12.166667 637.230 558.055 137.145 173.450
18840 37682 Bicycle 1097034 1097156 [585.15, 585.15, 585.145, 585.145, 585.145, 58... [171.855, 171.86, 171.86, 171.865, 171.865, 17... 123 4.100000 585.150 632.250 171.855 114.300
18841 37684 Bicycle 1097184 1097315 [636.625, 636.07, 635.52, 634.975, 634.435, 63... [127.15, 127.16499999999999, 127.18, 127.19999... 132 4.400000 636.625 567.120 127.150 163.760

11553 rows × 12 columns

In [3]:
df['coordinates'] = df.apply(lambda row: list(zip(row['xs_640x360'], row['ys_640x360'])), axis=1)
df.drop(columns=['xs_640x360', 'ys_640x360'], inplace=True)
df
Out[3]:
id class frame_in frame_out num_frames time_on_screen_s x_start_640x360 x_end_640x360 y_start_640x360 y_end_640x360 coordinates
0 2 Bicycle 250 409 160 5.333333 483.545 632.195 181.870 103.390 [(483.545, 181.87), (483.7, 181.93), (483.85, ...
3 8 Bicycle 250 398 149 4.966667 205.965 301.880 230.000 6.935 [(205.965, 230.0), (205.775, 229.27), (205.58,...
4 10 Bicycle 257 423 167 5.566667 606.620 297.715 145.220 280.260 [(606.62, 145.22), (602.955, 146.265), (599.28...
8 18 Bicycle 523 1482 960 32.000000 194.580 609.115 275.420 121.685 [(194.58, 275.42), (194.52, 275.18), (194.455,...
9 20 Bicycle 674 1655 982 32.733333 2.970 625.260 193.985 112.135 [(2.97, 193.985), (3.425, 193.64), (3.905, 193...
... ... ... ... ... ... ... ... ... ... ... ...
18837 37676 Bicycle 1096773 1096989 217 7.233333 625.370 307.340 138.795 276.935 [(625.37, 138.795), (622.415, 140.055), (619.4...
18838 37678 Bicycle 1096841 1096918 78 2.600000 545.280 456.185 187.275 231.440 [(545.28, 187.275), (543.88, 187.81), (542.495...
18839 37680 Bicycle 1096957 1097321 365 12.166667 637.230 558.055 137.145 173.450 [(637.23, 137.145), (636.435, 137.11), (635.65...
18840 37682 Bicycle 1097034 1097156 123 4.100000 585.150 632.250 171.855 114.300 [(585.15, 171.855), (585.15, 171.86), (585.145...
18841 37684 Bicycle 1097184 1097315 132 4.400000 636.625 567.120 127.150 163.760 [(636.625, 127.15), (636.07, 127.1649999999999...

11553 rows × 11 columns

In [4]:
df.frame_out.max()/60/1000
Out[4]:
18.288683333333335
In [5]:
def unixtime_to_datetime(unix_time):
    return datetime.fromtimestamp(unix_time)
 
def compute_datetime(row):
    # some educated guessing going on here: the paper states that the video covers 2021-06-09 07:00-08:00
    d = datetime(2021,6,9,7,0,0) + (row['frame_in'] + row['running_number']) * timedelta(seconds=2)
    return d

def create_point(xy):
    try: 
        return Point(xy)
    except TypeError:  # when there are nan values in the input data
        return None


new_df = df.head().explode('coordinates')
new_df['geometry'] = new_df['coordinates'].apply(create_point)
new_df['running_number'] = new_df.groupby('id').cumcount()
new_df['datetime'] = new_df.apply(compute_datetime, axis=1)
new_df.drop(columns=['coordinates', 'frame_in', 'running_number'], inplace=True)
new_df
Out[5]:
id class frame_out num_frames time_on_screen_s x_start_640x360 x_end_640x360 y_start_640x360 y_end_640x360 geometry datetime
0 2 Bicycle 409 160 5.333333 483.545 632.195 181.870 103.390 POINT (483.545 181.87) 2021-06-09 07:08:20
0 2 Bicycle 409 160 5.333333 483.545 632.195 181.870 103.390 POINT (483.7 181.93) 2021-06-09 07:08:22
0 2 Bicycle 409 160 5.333333 483.545 632.195 181.870 103.390 POINT (483.85 181.985) 2021-06-09 07:08:24
0 2 Bicycle 409 160 5.333333 483.545 632.195 181.870 103.390 POINT (484 182.045) 2021-06-09 07:08:26
0 2 Bicycle 409 160 5.333333 483.545 632.195 181.870 103.390 POINT (484.15 182.1) 2021-06-09 07:08:28
... ... ... ... ... ... ... ... ... ... ... ...
9 20 Bicycle 1655 982 32.733333 2.970 625.260 193.985 112.135 POINT (624.995 112.36500000000001) 2021-06-09 07:55:02
9 20 Bicycle 1655 982 32.733333 2.970 625.260 193.985 112.135 POINT (625.155 112.24000000000001) 2021-06-09 07:55:04
9 20 Bicycle 1655 982 32.733333 2.970 625.260 193.985 112.135 POINT (625.235 112.16499999999999) 2021-06-09 07:55:06
9 20 Bicycle 1655 982 32.733333 2.970 625.260 193.985 112.135 POINT (625.26 112.13499999999999) 2021-06-09 07:55:08
9 20 Bicycle 1655 982 32.733333 2.970 625.260 193.985 112.135 POINT (625.26 112.13499999999999) 2021-06-09 07:55:10

2418 rows × 11 columns

In [6]:
new_df.datetime.max()
Out[6]:
Timestamp('2021-06-09 07:55:10')
In [7]:
trajs = mpd.TrajectoryCollection(
    gpd.GeoDataFrame(new_df), 
    traj_id_col='id',  t='datetime', crs=None)
trajs
Out[7]:
TrajectoryCollection with 5 trajectories
In [8]:
trajs = mpd.MinTimeDeltaGeneralizer(trajs).generalize(tolerance=timedelta(seconds=10))
trajs.add_speed(overwrite=True)
traj = trajs.trajectories[2]
traj
Out[8]:
Trajectory 10 (2021-06-09 07:08:34 to 2021-06-09 07:14:06) | Size: 35 | Length: 338.2m
Bounds: (297.715, 145.22, 606.62, 280.26)
LINESTRING (606.62 145.22, 588.325 150.775, 570.99 158.285, 555.595 168.525, 540.77 178.485, 525.255
In [9]:
hvplot_defaults = {'line_width':5, 'frame_height':360, 'frame_width':640, 'colorbar':True, 'tiles':None, 'geo':False,}

traj.hvplot( **hvplot_defaults)
Out[9]:
In [10]:
bg_img = hv.RGB.load_image('img/intersection2.png', bounds=(0,0,640,360)) 
bg_img * traj.hvplot( **hvplot_defaults, c='speed', cmap='Reds')
Out[10]:
In [11]:
bg_img * trajs.hvplot( **hvplot_defaults, hover_cols=['id'])
Out[11]:
In [12]:
bg_img * trajs.hvplot( **hvplot_defaults, c='speed', cmap='Reds', hover_cols=['id'])
Out[12]:
In [13]:
import matplotlib.pyplot as plt
img_path = 'img/intersection2.png'
filtered_bg_img = plt.imread(img_path)
ax = plt.imshow(filtered_bg_img)
In [14]:
traj.plot()
Out[14]:
<Axes: >
In [15]:
traj.plot(ax=ax)

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
Cell In[15], line 1
----> 1 traj.plot(ax=ax)

File H:\miniconda3\envs\mpd-ex2\lib\site-packages\movingpandas\trajectory.py:242, in Trajectory.plot(self, *args, **kwargs)
    221 def plot(self, *args, **kwargs):
    222     """
    223     Generate a plot using GeoPandas default plotting (Matplotlib).
    224 
   (...)
    240     >>> trajectory.plot(column='speed', legend=True, figsize=(9,5))
    241     """
--> 242     return _TrajectoryPlotter(self, *args, **kwargs).plot()

File H:\miniconda3\envs\mpd-ex2\lib\site-packages\movingpandas\trajectory_plotter.py:154, in _TrajectoryPlotter.plot(self)
    152 if not self.ax:
    153     self.ax = plt.figure(figsize=self.figsize).add_subplot(1, 1, 1)
--> 154 ax = self._plot_trajectory(self.data)
    155 self.kwargs[
    156     "legend"
    157 ] = False  # has to be removed after the first iteration, otherwise we get multiple legends!  # noqa E501
    158 self.kwargs.pop(
    159     "column", None
    160 )  # has to be popped, otherwise there's an error in the following plot call

File H:\miniconda3\envs\mpd-ex2\lib\site-packages\movingpandas\trajectory_plotter.py:80, in _TrajectoryPlotter._plot_trajectory(self, traj)
     78     self.kwargs.pop("vmin", None)
     79     self.kwargs.pop("vmax", None)
---> 80     line_plot = line_df.plot(
     81         ax=self.ax,
     82         vmin=self.min_value,
     83         vmax=self.max_value,
     84         *self.args,
     85         **self.kwargs
     86     )
     87 return line_plot

File H:\miniconda3\envs\mpd-ex2\lib\site-packages\geopandas\plotting.py:969, in GeoplotAccessor.__call__(self, *args, **kwargs)
    967 kind = kwargs.pop("kind", "geo")
    968 if kind == "geo":
--> 969     return plot_dataframe(data, *args, **kwargs)
    970 if kind in self._pandas_kinds:
    971     # Access pandas plots
    972     return PlotAccessor(data)(kind=kind, **kwargs)

File H:\miniconda3\envs\mpd-ex2\lib\site-packages\geopandas\plotting.py:685, in plot_dataframe(df, column, cmap, color, ax, cax, categorical, legend, scheme, k, vmin, vmax, markersize, figsize, legend_kwds, categories, classification_kwds, missing_kwds, aspect, **style_kwds)
    681         ax.set_aspect(1 / np.cos(y_coord * np.pi / 180))
    682         # formula ported from R package sp
    683         # https://github.com/edzer/sp/blob/master/R/mapasp.R
    684     else:
--> 685         ax.set_aspect("equal")
    686 elif aspect is not None:
    687     ax.set_aspect(aspect)

AttributeError: 'AxesImage' object has no attribute 'set_aspect'
In [ ]: